Stanley Kubrick's 1968 film, "2001: A Space Odyssey," showed a future that seemed distant enough to give us time to invent it. Sure enough (despite laggard progress into space), we've already grown used to everyday encounters with many of the movie's once-futuristic concepts, from handheld devices to supercomputers.

In the spirit of the holistic vision of Kubrick and collaborator Arthur C. Clarke, we find that no single development dominates eWEEK Labs' expectations for IT innovation in 2001. We aren't even tempted to propose a facile label such as "Year of Wireless" or "Year of Convergence," though many such labels will surely vie for enterprise mind share.

Instead, we see continued impressive progress on many complementary fronts in this 2001 tech odyssey.

Growing network capacity will invite richer media streams and enable more complex transactions. Business and entertainment activities will generate floods of records and will demand the creation and distribution of oceans of data, requiring vast storage investments (with 64-bit systems to track it all). Wireless connectivity and growing network dependence will elevate concerns regarding integrity and privacy.

The technologies that we examine in this special report solve pressing enterprise problems, create new problems of their own and play cooperative roles in curing each other's side effects. We hope you'll find this a useful watch list, as well as a wish list, as you enter the 21st century.

Tech item 1: AES
A code-breaking scheme that takes only 1 second to defeat today's DES would need 149 trillion years to crack a 128-bit implementation of the forthcoming AES. But even as strong encryption becomes the immune system of both e-businesses and personal transactions, every enterprise IT decision maker should understand the many factorsââ,¬"not just strengthââ,¬"that went into selecting the proposed Advanced Encryption Standard algorithm called Rijndael.

On strength alone, Rijndael was not even the leading contender among the five finalists for adoption as AES, but the mathematics of all modern encryption schemes are robust to the point of overkill.

In most cases, the chance that hardware costs or software overheads will lead to weak implementations is a far greater risk than cryptanalytic attacks. Rijndael has an important edge, therefore, with its economical use of both memory and processing power.

The royalty-free Twofish algorithm (also an AES candidate), more secure and nearly as economical, will be protecting many users through its incorporation into renowned cryptographer and privacy advocate Phil Zimmerman's PGP 7.0 (which Zimmerman now plans to extend with a Rijn dael option, because AES adoption means that Rijndael must also become freely available). Major enterprisesââ,¬"and perhaps national governmentsââ,¬"may prefer to license the computationally extravagant but apparently airtight MARS.

All crypto users, however, must recognize that protection is not so much a question of "how strong" as "how long." If data has only transient value, the present DES (Data Encryption Standard) or the widely used Triple DES may still suffice; if privacy is desired for decades after interception of encrypted data, 256-bit keys and other AES options belong in the arsenal of enterprise data defenders.

ââ,¬"Peter Coffee

2. Bandwidth management and QOS

The problem of managing QOSââ,¬"the prioritised allocation of network bandwidth when transmissions contend for pipeline positionââ,¬"will blossom next year as IT managers face ever-greedier applications and an increase in raw traffic volume.

Even with the increased importance of quality of service and its cousins, SLA (service-level agreement) and class of service, eWeek Labs does not foresee any meteoric advances next year. The Common Open Policy Service technology endorsed by Cisco Systems Inc. should evolve a bit, and the Distributed Management Task Force will likely make headway on QOS-related standards. Elsewhere, there's not much brewing.

Rather than scanning the horizon for technological advances, IT managers should focus on relationships with service providers and haggle to get traffic prioritised. With the unsettled state of the provider market, the time is right to demand lower rates and forge favorable SLAs.

Bandwidth management will be a critical issue next year for companies adopting VOIP (voice over IP) or other technologies that require timely delivery of data. Managers of traditional data-only networks should leave QOS to those on the bleeding edge of VOIP and focus, instead, on providing sufficient capacity to handle the heaviest consistent loads of data traffic.

One problem that won't go away in 2001 is the complexity of implementing QOS, especially in heterogeneous networks. Network equipment vendors seem almost to take delight in blithely ignoring customers' needs for interoperability among devices from diverse vendors. Managers should look to products such as IPHighway Inc.'s Open Policy System to help set policies across disparate equipment.

Similarly, upcoming technology changes such as IPv6 will likely make overall network management easier by including QOS as well as IP Security in the core of the TCP/IP protocol.

3. Data mining

In 2001, lower storage costs and a growing variety of useful data analysis tools and services will continue to glean more value from data being mined.

The growth of e-business has created a data overload for many companies, and sorting through it is becoming harder. Through centralized data management and analytical solutions, data mining will play an integral role in all businesses focused on customer relationship management and enterprise resource planning.

A number of vendors offer data mining tools that can handle any company's needs. Vendors such as IBM, Smart Research BV and Blue Martini Software, among a host of others, offer software solutions or data mining engines to environments ranging from Windows to Linux and Unix.

Storage of all that information is becoming less of a headache. Online analytical processing-based data warehousing and services have addressed the critical problems of what to store and where to put it. Companies that have been hesitant to integrate young, untested technology will have more options as the technologies mature next year and beyond.

More flexible tools and warehousing packages will make data mining a category to watch in 2001.

4. JavaServer Pages

Java has established itself as the dominant language for server-side programming. Through JSP, Java is going to become equally important in 2001 as a Web application development language.

Java makes it easier for developers to write safe, reliable programs through features such as automatic memory management and structured exception handling, while its large set of APIs and cross-platform design provide power and great portability.

JavaServer Pages provides a seamless way to connect Web applications, which are usually written in HTML and by Web designersââ,¬"not professional programmersââ,¬"with the growing assortment of back-end business logic written in Java. No other Web scripting language provides as consistent and as integrated a development model for both Web application and back-end logic development. (When Microsoft releases its .Net framework in 2001, the company will provide a similarly comprehensive Web programming model.)

However, JSP has had a few years on the market to become understood and to mature. The technology will hit critical mass in 2001.

Although the specification is just at Version 1.1, JSP is already widely supported among high-end application servers and on platforms ranging from PCs to mainframes, because JSP support is a required part of Sun Microsystems' Java2 Enterprise Edition Java application server specification.

On the low end, the Apache Group has released Tomcat, a free, open-source JSP engine that is itself written in Java and so can be run anywhere Java runs. Tomcat's liberal licensing terms make it attractive and easy for software vendors to embed Tomcat in their applications.

JSP is unique among Web scripting languages in its support for both a tag-based programming style (such as those that Allaire's Cold Fusion made popular), using tag libraries, and an embedded script-block programming style (such as Microsoft's Active Server Pages or the PHP Development Team's PHP use).

A relative lack of integrated development environments has been an obstacle for JSP, but this year saw the introduction of several new tools (ranging from Macromedia's Dreamweaver UltraDev to Allaire's JRun Studio to Web Gain's WebGain Studio) that target JSP developers.

5. Middleware

The popularity of handheld computers with wireless connectivity continues to grow, but the effectiveness of these devices in the enterprise space will remain limited until companies implement middleware technologies that enable these devices to access critical data and applications on the corporate network.

Currently, many corporate users of handheld devices are forced to access e-mail via messy forwarding schemes and are able to wirelessly access little more than sports scores and daily horoscopes.

With an eye toward brightening the prospects of handheld computers in the enterprise, Palm, Microsoft, and a host of software vendors have begun to offer middleware products that sit between end-user devices and the servers that store the data that mobile workers must access.

It's important that the middleware products that companies deploy include support for multiple mobile device platforms. Palm is the current leader in market share, but Windows CE-based devices are maturing nicely, and upcoming wireless-enabled devices from Symbian figure to make a large impact once they reach the market next year.

Although many of these middleware products comply with Internet standards to ensure that they work with more than one vendor's mobile device, the best products will also include client-side technology on the handheld to handle offline use. This is a vital feature because with mobile devices, the connection to the Net can be unreliable.

These middleware products will generally be available both in self- hosted and application service provider models; in the coming year, enterprise IT managers should determine which model works for their company, based on available IT resources.

6. PCIX

Managers of enterprise networks should get help increasing bandwidth and minimising performance bottlenecks in 2001 from the new PCIX bus architecture, which has the potential to dramatically increase I/O performance in enterprise servers.

Since its development in the early '90s, the PCI local bus has been a fundamental component in most computer systems. It replaced the legacy ISA bus as the de facto peripheral bus in all systems, from home computers to enterprise servers.

Most PCI buses now in use are 32-bit and limited to speeds of 33MHz and 133M-bps throughput. Faster enterprise servers are built with the 66MHz PCI buses with 64 bits of bandwidth, but even those servers can't keep pace with the components around them. The latest processors handle transmissions considerably faster than a 66MHz PCI bus can, as do peripheral devices based on Fibre Channel, Gigabit Ethernet or Ultra160 SCSI.

The proliferation of distributed computing is also driving demand for a faster bus architecture. As enterprise sites add clustered server farms and shared storage with switched-fabric storage area networks to meet demand for greater scalability and higher availability, bus speed becomes critically important.

To break the current PCI speed barrier, Compaq, Hewlett-Packard and IBM joined forces in 1998 to develop the 64-bit PCIX bus architecture, which will run at 133MHz and attain throughput of 1,066M bps.

An important aspect of the new PCIX bus architecture is backward compatibility with the current PCI standard. A PCI adapter will function in a PCIX system and vice versa, but a PCIX adapter used in a PCI bus will be limited to standard PCI speeds.

The PCI Special Interest Group approved the PCIX standard in September 1999, and vendors are expected to ship the first PCIX-enabled product at the end of this year or early next year.

Although it will improve throughput significantly, the PCIX architecture will eventually be replaced by the InfiniBand interconnect architecture that is now in development. With InfiniBand-based products still two to three years away, the HAL 9000 supercomputer will likely miss its 2001 ship date, but the PCIX bus will become part of the IT landscape.

7. XML Schema

In the four years since it was released, XML has become widely used in a variety of applications and systems employed today in e-business. Extensible Markup Language is also a core technology in Microsoft's linchpin .Net initiative. However, XML has yet to reach its greatest expectation of being the ubiquitous language used to enable business-to-business data communications.

This should finally start to change early in 2001, when the World Wide Web Consortium releases XML Schema as a recommendation, which will essentially make it a standard.

The main problem holding XML back as a mechanism for B2B communication has been that systems using XML created for one company or industry cannot understand XML documents created for another company or industry. This is because each one is using a different Document Type Definition or schema, and the systems have no starting point for deciphering differences.

XML Schema will change this. Simply put, it will bring rich data descriptions to XML. By defining a set of shared markup vocabularies, it will provide a method for describing all the data in an XML file. Businesses and industries will still be able to create the specific schema information needed for their market but, because they will be starting from a common base, that information will be much easier for other systems to understand.

There are several other schema definitions being used by various industries and groups, including the Microsoft-backed BizTalk framework. However, most groups have publicly stated that they will move to support and incorporate the W3C's XML Schema.

Like XML itself, it will take some time before XML Schema is in everyday use. However, its release will finally allow businesses to start building XML integration. And it should also allow businesses to take a look at several other XML-based standards that really need XML Schema as an underpinning, such as Resource Description Framework.

8. Optical switching

Let's start with a pretty safe prediction: The Internet isn't going to get any smaller or less popular in 2001. Staying on a roll here, let's add this one: Keeping Internet data moving fast and reliably to and from business partners and customers will be a key IT objective.

One of the more promising technologies to make sure that Internet and WAN links continue to scale with demand is all-optical (photonic) switching, where only at the end points of a communication link is the communication signal electrical.

End-to-end optical links are still some years off because of the low cost of copper wiring and network cards vs. fibre wiring and fibre-capable cards, but on backbones, all-optical switches and the amazing micromirror arrays that power them have already become a reality in products such as Lucent Technologies's WaveStar Lambda Router.

There are still many significant problems with all-optical switching, however. Higher-level network layer switching, which requires decoding inner layers of packet data to get routing information, remains tricky, as does the need to regenerate signals as they degrade, which requires a conversion to electrical signals and then an optical retransmission.

Still, there has been significant progress in this area, and the frequency density of fibre transmission techniques keeps rising, offering more flexibility and ever-greater throughput on optical links.

Moving from macroscopic to microscopic technology, optical transistors (which are a key step to fully optical microprocessors) are also getting closer to reality. However, we are still years away from high-volume, high-yield manufacturing processes for these components. Early research holds the promise of a hundredfold speedup over current electrical transistors, so the potential of all-optical computing remains tantalising.

9. 64-bit processors

Intel faces a big job trying to lure customers away from current 64-bit processors, but Itanium, the company's first 64-bit processor, will be the one to watch in 2001.

Intel hopes to use its considerable industry muscle to supplant entrenched RISC-based systems sold by competitors such as Sun Microsystems and Compaq with the EPIC (Explicitly Parallel Instruction Computing) Itanium, which was co-designed by Hewlett-Packard.

eWEEK Labs believes that the introduction of Itanium will, if nothing else, help drive down the steep prices for 64-bit RISC-based systems.

The evolution of the EPIC IA-64 processor depends largely on developments in compiler technology. We expect compiler-related improvements to come slowly, however, because ISVs face a long learning curve in adapting to the IA-64 architecture.

Itanium will be focused toward the high-end server market, in particular the database sector, where Itanium will expand memory capacity that in most cases now maxes out at 4GB for 32-bit processors.

An important secondary market will be in the workstation space, where systems will be able to use the Itanium's 64-bit power to solve complex computations.

The strong support for Itanium that Intel enjoys from virtually every major vendor of hardware and operating systems should foster the acceptance of Itanium-based servers and workstations.

IT managers contemplating the jump to 64-bit computing should also evaluate the latest operating systems before making their platform decisions. Several 64-bit systems, including Microsoft's Whistler, Novell's Modesto and 64-bit Linux, are expected to ship with Itanium-based systems next year.

10. Linux 2.4 Kernel

Linux will remain a key technology to watch next year, and the action should begin in the first quarter following the release of Version 2.4 of the Linux kernel, which should usher in significant advances. Linux creator Linus Torvalds has said the 2.4 kernel is likely to become available this month, so the first 2.4-based Linux distributions should ship in the first quarter of next year.

Linux has enjoyed a highly successful run since the current 2.2 kernel shipped in January 1999. Vendors of enterprise-critical server applicationsââ,¬"high-end databases, application servers, Web storefronts, firewalls and suchââ,¬"have embraced Linux en masse. Most large server manufacturers now offer and support Linux on their hardware. Likewise, vendors of network appliances have made generous use of the low-cost, highly customisable and highly reliable operating system.

Like the sellers, buyers in corporate IT have migrated to Linux-based servers of many types, including Web, Domain Name System, file and print, Internet access, and e-mail. Linux is particularly popular for departmental use, where loads are relatively light and management is handled remotely.

The Linux 2.4 kernel development process has focused on improving performance on larger machines and building in support for new hardware options such as Universal Serial Bus and architectures such as Intel's 64-bit Itanium and IBM's S/390.

Another important goal has been to improve SMP (symmetric multiprocessing) performance. A new process-scheduling algorithm to speed response time for network server applications such as Web servers is in the works, as are an SMP-improved disk cache design and a more SMP-capable networking layer. Developers have targeted Web server performance with a kernel-level Web server called khttpd. Another new Linux kernel-level Web server, Red Hat's Tux, posted spectacular results this year in Standard Performance Evaluation's SPECweb99 Web server benchmark.

A journaling file system won't make it into the initial release of Linux 2.4, but several efforts are close to fruition, and Linux will likely gain this crucial reliability feature sometime next year. Support for files larger than 2GB is in this kernelââ,¬"a welcome change, especially for systems with large log files and database servers.

Next year will be an important time for Linux in other ways, too. Inprise will ship its Kylix development tool, and Sun Microsystems is finally providing the same level of Java support on Linux that it delivered for other platforms.

On the desktop, Linux's market share remains very small, but integrated desktop and application suites such as K Desktop Environment and GNU Network Object Model Environ ment are improving rapidly.

11 & 12. Voice over IP, Storage over IP

Voice over IP
Going into 2001, eWEEK labs expects to see a variety of advances in voice-over-IP technology. Implementations are another matter, however, because at the end of the day, vendors of VOIP wares will still have trouble countering the old saying, "If it ain't broke, don't fix it."

Despite the loud buzz that has surrounded VOIP for close to three years, most major corporations are still waiting and watching. They demand high-quality phone service and take it for granted because the technology is mature and reliable.

Those adjectives still don't come to mind when one thinks of Internet service providers and network applications, and most companies won't risk the quality of a service as basic and mission-critical as telephony.

With the advent of supporting applications such as unified messaging, VOIP will gain credibility and might even fulfill vendors' claims that it is destined to be a must-have technology. But 2001 will not be the year when most major corporations replace their telephony infrastructures with VOIP systems.

A key area to watch next year will be the development of chip and codec products from companies such as Lucent Technologies Inc. and Global IP Sound. Those products will allow vendors to improve the sound quality of IP calls while making network utilization more efficient.

Another critical area to watch is the evolution of soft switches (software switches). Soft switches are responsible for call-control functions such as call routing, admission control and connection control. 3Com and Lucent will likely peddle their soft-switch products vigorously in 2001.

Last but certainly not least, IT managers should watch the headlines for hints of heavy-handed government regulation that could pour cold water on the entire VOIP movement while protecting Old Guard telephone service providers.

ââ,¬"Henry Baltazar

Storage over IP
As IP networks grow to carry ever more enterprise data, the demand for storage and access to archives grows in (at least) equal proportion. Economics will make IP storage connection an important item on the 2001 agenda.

We'd like to call this concept "storage over IP," or SoIP, by analogy to voice over IP; unfortunately, the attractive "SoIP" label is a copyrighted term owned by Nishan Systems, pushing the industry toward some less convenient label such as "SANoIP" for the next step beyond today's Fibre Channel SANs.

Worldwide capacity on IP is growing, performance and cost of IP switching equipment are improving, and quality of service on IP is being propelled by the stringent demands of streaming media. For enterprise data administrators, SANoIP has the further attraction of consolidating network administration and other technical support around a core set of knowledge, skills and tools.

Administrators should be wary, however, of letting expectations get too high. Distributed systems, it's been said, work best when they're no more than five feet apart: It's useful to maintain short signal paths between storage systems that have to agree with each other. Longer paths suffer from speed-of-light and switching delays that threaten cache coherence. They also suffer from risks of physical interruption.

Prosaic problems such as cable connector types, and esoteric problems such as autonegotiation standards for heterogeneous network interactions, can create unforeseen costs and prolong adoption timetables. But SANoIP adoption can be incremental: Initial investments can focus on connection of dispersed Fibre Channel SANs rather than relying on bleeding-edge technology for their most critical systems.

ââ,¬"Peter Coffee

Copyright © 2009 CBS Interactive, a CBS Company. All Rights Reserved.
ZDNET is a registered service mark of CBS Interactive. ZDNET Logo is a service mark of CBS Interactive.